Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Contents lists available at ScienceDirect

Int J Appl Earth Obs Geoinformation

journal homepage: www.elsevier.com/locate/jag

Assessing effects of urban vegetation height on land surface temperature in T

the City of Tampa, Florida, USA
⁎
Qiuyan Yua, , Michael Acheampongb, Ruiliang Pua, Shawn M. Landrya, Wenjie Jic,
Thilanki Dahigamuwad
a
School of Geosciences, University of South Florida, Tampa FL 33620, USA
b
Department of Geography, Oklahoma State University, Stillwater OK 73107, USA
c
Plant and Environmental Sciences, New Mexico State University, Las Cruces NM 88003, USA
d
Department of Civil and Environmental Engineering, University of South Florida, Tampa FL 33620, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Urban vegetation can mitigate urban heat island (UHI) due to its ability to regulate temperature by directly or
Urban heat island (UHI) indirectly influencing water vapor transport, shading effect, and wind speed and direction. Mechanisms of effects
Land surface temperature (LST) of vegetation cover on land surface temperature (LST) have been extensively documented. Few studies, however,
Vegetation fraction have examined the role of vegetation height in controlling LST. In this study, we examined the relationship
Vegetation height
between LST and vegetation height by using Light Detection and Range (LiDAR) data from the city of Tampa,
LiDAR
Florida, USA. The results revealed that vegetation height has significant impact on LST. Additionally, we also
identified the optimal height and fractional cover at which vegetation can exert the greatest influence on LST. In
particular, we found that the maximum cooling effect of vegetation can only be achieved when vegetation cover
is above 93.33%, an amount of which is nearly impossible to have in most of the cities. On the other hand, LST
decreases at an increasing rate with vegetation height, and is optimized at 20 m. This shows that vegetation
height can play an important role in regulating UHI in contributing to effect maximization with least cover
possible in a city. Findings derived from this study could provide urban planners with critical insights on precise
and efficient urban vegetation management in the purpose of UHI mitigation.

1. Introduction Therefore, mitigating UHI effects, especially under climate change, is

necessary for the promotion of urban sustainability (Akbari et al.,
Urban warming has become a serious problem due to global change 2001).
and urbanization. One important phenomenon is the increasing urban City-level adaptation strategies to mitigate UHI have important
heat island (UHI) effect, which can increase water consumption and economic and environmental benefits for most of the cities around the
energy use (Akbari et al., 2001; White et al., 2002), elevate environ- world. Planting of vegetation in urban areas is one of the most widely
mental pollution (Stone, 2005), and compromise human health and applied methods (Bolund and Hunhammar, 1999; Susca et al., 2011).
comfort (Tan et al., 2010). With the majority of the world’s population The structure of vegetation affects temperature by directly or indirectly
residing in cities, it is important and urgent to examine whether and influencing water vapor transport, shading effect, and wind speed and
how urban regions can become more sustainable under global change direction. The two fundamental aspects of vegetation structure are
(Campbell, 1996). UHI affects the sustainability of urban regions and horizontal and vertical structure. Attributes of the horizontal structure
residents worldwide (Murgante et al., 2011). About 60% of the global of vegetation such as vegetation cover, crown size, and crown shape all
urban population is currently experiencing twice as much warming as contribute to evaporative fraction and turbulent fluxes (Su et al., 2001).
the world’s average and the temperature in cities is projected to in- Height, as another major attribute of vegetation structure, and a pri-
crease by another 2 °C by 2050 (Johnston, 2017). Any hard-won vic- mary determinant of surface roughness, has a significant influence on
tories over climate change on a global scale could be wiped out by the water vapor transport and convergence, wind speed, and shading area
effects of uncontrolled UHI (Estrada et al., 2017; Hakner, 2017). (Geier-Hayes et al., 1995; Maurer et al., 2015; Su et al., 2001; Sud et al.,

⁎
Corresponding author.
E-mail addresses: [email protected] (Q. Yu), [email protected] (M. Acheampong), [email protected] (R. Pu), [email protected] (S.M. Landry),
[email protected] (W. Ji), [email protected] (T. Dahigamuwa).

https://doi.org/10.1016/j.jag.2018.08.016
Received 4 April 2018; Received in revised form 8 August 2018; Accepted 13 August 2018
0303-2434/ © 2018 Elsevier B.V. All rights reserved.
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

1988). Therefore, understanding how vegetation height affects UHI All remote sensing data were acquired in 2006 and the difference of
would contribute to precise urban greenspace management approaches months between the LiDAR data and the rest of the data would not be
to alleviate UHI effects. expected to have a meaningful impact on the analysis, as the vegetation
Considerable number of studies have demonstrated the significance height should not change much during this time period in Tampa.
of the horizontal structure of urban vegetation in decreasing tempera-
ture (Alkama and Cescatti, 2016; Declet-Barreto et al., 2016; Deilami 3. Methodology
et al., 2018; Susca et al., 2011; Weng et al., 2004; Wu et al., 2014). It is
obvious that the greater the extent of vegetation cover is, the stronger Data analysis includes two main procedures: (1) retrieving LST,
the cooling effect would be. However, since cities have limited space to vegetation fraction, and mean vegetation height, and (2) exploring the
extend vegetation cover, reducing UHI effect by increasing vegetation relationships between LST and mean vegetation height (Fig. 4).
cover might be impossible in many cities. Moreover, height is an at-
tribute that is easier and less costly to manage (e.g. through pruning 3.1. LST retrieval using the radiative transfer equation (RTE) algorithm
and selection of tree species) than vegetation cover. A better under-
standing in the relationship between vegetation height and LST would We calculated LST using the radiative transfer equation (RTE) al-
provide significant insight to efficient UHI mitigation. Contrary to re- gorithm (Fu and Weng, 2016; Jiménez‐Muñoz and Jiménez‐Muñoz and
search on the effects of vegetation cover, the impacts of height on urban Sobrino, 2006; Sobrino et al., 2004; Weng and Fu, 2014). According to
temperature have rarely been studied. A recent study conducted in the RTE algorithm, TIR radiance at wavelength λ can be expressed as
Chicago, Illinois found that vegetation volume affects summer night- (Sobrino et al., 2004):
time temperature (Davis et al., 2016). Using the City of Tampa, Florida
↓ ↑
as an example, this study aims to investigate the effects of vegetation Lsensor , λ = [ελ Bλ (Ts ) + (1−ελ ) Latm , λ ] τλ + Latm, λ (1)
height on LST so that it can be used to combat UHI effects and develop
where ελ is land surface emissivity (LSE), Bλ(Ts) is blackbody radiance
precise urban greenspace management plans.
, λ , Latm, λ , and τλ are atmo-
↓ ↑
at land surface temperature Ts (K), and Latm
spheric parameters representing the downwelling atmospheric ra-
2. Study area and data diance, the upwelling atmospheric radiance, and the total atmospheric
transmissivity between the surface and the sensor, respectively. The
The study area is located in the City of Tampa, Florida (28 °N,
, λ , Latm, λ , and τλ for each Landsat 5 T M
↓ ↑
atmospheric parameters Latm
82 °W), with a total area of 350 km2 (Fig. 1). The region is characterized scene were obtained from the NASA Atmospheric Correction Parameter
by a humid, sub-tropical climate with warm, wet summers and mild, Calculator (Barsi et al., 2003, 2005). The LST can then be derived from
dry winters. The extent of urban land use has increased approximately Bλ(Ts) using Planck’s law:
sevenfold during the last century in the Tampa Bay area (Xian and
C1
Crane, 2003, 2005). Due to rapid growth of the city over the past 40 Bλ (Ts ) =
years, impervious surface has expanded extensively into exurban zones
and native forests (Xian and Crane, 2005), resulting in a heterogeneous
λ5 ⎡exp
⎢
⎣
( )−1⎤⎥⎦
C2
λTs
(2)
landscape of impervious surface and urban vegetation. By 2006, the −1
where C1 and C2 are constants (C1 = 1.191 × 10 Wμm sr m-2, 8 4
City of Tampa was comprised of 36% impervious surface, 58% vege-
C2 = 1.439 × 104μm·K).
tation (of which 29% is tree canopy), 2% water, and 4% bare land/soil
LSE (ελ) was estimated using the NDVI Thresholds Method
(Landry and Pu, 2010).
(NDVITHM) proposed by Sobrino et al. (2008). NDVI was calculated
Datasets used in the study included Landsat 5 thematic mapper
from atmospherically-corrected red and near infrared bands of the
(TM) land surface reflectance (LSR) product, Landsat TM level-1
Landsat 5 T M LSR product. The NDVITHM method assigns different
thermal infrared (TIR) product, Light Detection and Ranging (LiDAR),
emissivity to bare soil (NDVI < 0.2), full vegetation (NDVI > 0.5),
and high-resolution land cover map (Table 1). Landsat 5 T M products
and soil-vegetation-mixed pixels (0.2 ≤ NDVI ≤ 0.5):
(both LSR and TIR) acquired on May 2, 2006 were downloaded from
the USGS EarthExplorer website (https://earthexplorer.usgs.gov). The ⎧ 0.97, NDVI < 0.2
LSR images were derived after applying the atmospheric correction ελ = 0.004Pv + 0.986, 0.2 ≤ NDVI ≤ 0.5
⎨
routines of MODIS (the Second Simulation of a Satellite Signal in the ⎩ 0.99, NDVI > 0.5 (3)
Solar Spectrum, done by U.S. Geological Survey (USGS)) to the original
where Pv is the fraction of vegetation and according to Carlson and
Landsat Thematic Mapper (TM) images. The resulting pixel value of the
Ripley (1997) can be estimated as:
LSR data, ranging from 0 to 1, is a standardized reflectance factor of the
land surfaces. The TIR product provides thermal infrared data at wa- 2
NDVI −NDVImin ⎞
velengths between 10.4 μm to 12.5 μm and was used to derive LST. Pv = ⎛ ⎜ ⎟

⎝ NDVImax −NDVImin ⎠ (4)

LiDAR data provided by Southwest Florida Water Management District
were used to obtain mean vegetation height. LiDAR data was collected where NDVImin and NDVImax are 0.2 and 0.5, respectively.
by Optech Gemini scanners at a maximum frequency of 167 kHz with
four returns and an average point density of 1.51 points per m2. The 3.2. Extraction of vegetation fraction and mean vegetation height
first and last return of LiDAR data were used to produce digital surface
models (DSM) and digital elevation models (DEM), respectively, using a Vegetation fraction and mean vegetation height were extracted by
binning average interpolation technique (1 m cell resolution). Nor- zoning the vegetation cover map and the nDSM of vegetation to the
malized digital surface model (nDSM), derived by subtracting the DEM same spatial resolution of the LST image (30 m). Vegetation cover map
form the DSM, is used to model the above ground height of vegetation was derived by selecting tree canopy and other vegetation from the
(Figs. 2 and 3). Land cover was mapped using object-based artificial land cover map, and then vegetation fraction was assigned as the sum of
neural network from the pan-sharpened IKONOS 1 m panchromatic and all cell values in each zone divided by total number of cells in each
4 m multispectral bands) acquired in April 6, 2006 (Landry and Pu, zone. The nDSM of vegetation was obtained by extracting nDSM cov-
2010). Vegetation in the land cover map includes tree canopy and other ered only by vegetation, and the mean vegetation height was calculated
vegetation (such as grass, lawn, and shrub). All the datasets were ac- as the mean of all cell values in each zone. Zoning vegetation cover and
quired under a clear sky condition and projected to the same coordinate nDSM of vegetation using the pixels of LST can ensure LST, vegetation
system (Florida West State Plane NAD83) and subset to the study area. fraction, and mean vegetation height to be geo-coregistered. Vegetation

713
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Fig. 1. The City of Tampa, Florida, USA. World topography map sources: Esri, HERE, Garmin, Intermap, INCREMENT P, GEBCO, USGS, FAO, NPS, NRCAN, GeoBase,
IGN, Kadaster NL, Ordnance Survey, Esri Japan, METI, Esri China (Hong Kong), swisstopo, MapmyIndia, © OpenStreetMap contributors, GIS User Community.

Table 1
Data description.
Description Acquisition Date Data source Spatial resolution
(m)

Land Surface Reflectance LSR image includes blue, green, red, near infrared, and shortwave May 2, 2006 USGS 30
(LSR) product infrared bands.
Thermal infrared (TIR) The TIR image was resampled from 120 m to 30 m by U.S. Geological May 2, 2006 USGS 30
product Survey (USGS).
Land cover Land cover map was obtained from the pan-sharpened IKONOS 1 m April 6, 2006 Landry and Pu (2010) 1
panchromatic and 4 m multispectral bands. Land cover types include
tree canopy, other vegetation, bare sand/soil, water, and impervious
surface.
Normalized Digital Surface Obtained by subtracting the digital elevation model from the digital January to Southwest Florida Water 1
Model (nDSM) surface model. February, 2006 Management District

714
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Fig. 2. The workflow of retrieving the normalized digital surface model (nDSM). Acronyms in this figure: DSM − digital surface model, DEM − digital elevation
model, nDSM−normalized digital surface model.

fraction value at each pixel stands for the fractional cover of vegetation 3.3. Statistical analyses
including tree canopy and other vegetation. Mean vegetation height
representing a vertical center of a canopy’s volumetric profile has been A Pearson correlation matrix was used to examine the strength of
used in this study to represent vegetation height. It has been frequently the association between LST and mean vegetation height. For com-
used to represent plot-level vegetation height (Asner et al., 2011; Asner parison, the correlation coefficient of LST with vegetation fraction was
and Mascaro, 2014). also calculated. We also used multiple-linear regression analysis, where
standardized coefficients (β) were used to compare the strength of the
effect of vegetation fraction and height in predicting LST (Weng et al.,

Fig. 3. Sample images showing (a) a subset of the study area, (b) the land cover map, and (c) normalized digital surface model (nDSM). The height ranges from 0 m
(blue) to 28 m (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

715
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Fig. 4. The workflow of examining the relationships between vegetation structure and land surface temperature (LST): (a) examining the correlations of LST with
vegetation fraction and height, and (b) investigating the effects of variance of vegetation fraction and height on LST. Acronyms in this figure: nDSM − normalized
digital surface model, LST–land surface temperature, ANOVA–analysis of variance.

2006; Yan et al., 2014; Zhou et al., 2011, 2017). The higher the abso- MacDill Air Force Base. The overall mean vegetation height in the study
lute value of the standardized coefficient is, the stronger the effect is. area was approximately 5.14 m. Most of the taller trees (> 10 m) were
We further used one-way analysis of variance (ANOVA) and esti- located in the preserved area in the northeast (Fig. 5(d)).
mated marginal means to examine the effects of variance of mean ve-
getation height on LST. One-way ANOVA was used to test whether the
4.2. Correlation of LST with vegetation fraction and mean vegetation height
means of LST are significantly different between different ranges of
vegetation cover and height. To achieve this, we grouped vegetation
Results of Pearson correlation analysis showed that both vegetation
fraction and height into 15 groups. After trying different groupings, we
fraction and vegetation height were negatively correlated with LST.
chose 15 groups, as they had the lowest intra-group variance, but
Vegetation height had a higher correlation with LST (correlation coef-
highest inter-group variance. The analysis was been done by SPSS (IBM
ficient = −0.579) compared to vegetation fraction (correlation coeffi-
Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0.
cient = −0.479). The standardized coefficients (β) of multiple regres-
Armonk, NY: IBM Corp.)
sion analysis also indicated mean vegetation height (β = −0.235) as a
stronger predictor of LST than vegetation fraction (β = −0.452)
(Table 2).
4. Results

4.1. LST, vegetation fraction, and height 4.3. Effects of vegetation fraction and mean vegetation height on LST

LST ranged from 292.40 K (19.25 °C) to 324.15 K (51.00 °C) in the One-way ANOVA analysis showed a statistically significant differ-
City of Tampa (Fig. 5(b)). UHIs can be observed in the downtown area ence in mean LST between the different levels of vegetation fraction
(close to south of Tampa), near University of South Florida, around the (p < 0.05). Mean LST decreases with increasing vegetation fraction.
two airports (Tampa International Airport and MacDill Air Force Base; Noticeably, the estimated marginal means of LST decreases at a rela-
south of Tampa; Fig. 5(a) and (b)), and along the main roads. The tively consistent rate when vegetation fraction is below 93.33%, while
“oases”, where LST was about 20 K less than the UHIs, are located in it decreases sharply when vegetation fraction is above 93.33% (Fig. 6).
south Tampa near the water, the neighborhood parks, and the natural The result indicated that the greatest amount of cooling effect was
reserves in the north. Spatial pattern of vegetation corresponds well to achieved when vegetation fraction was higher than ∼93.33%.
LST, where cover was highest in the north and lowest in the south On the other hand, significant difference in mean LST was also
(Fig. 5(c)). Other small clusters of vegetation were also found along the found between different levels of mean vegetation height (p < 0.05)
Hillsborough River and around the southern peninsula area near the for vegetation height less than 20 m. LST decreases at an increasing rate

716
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Fig. 5. (a) points of interest of the study area, (b) land surface temperature (LST), (c) vegetation fraction, and (d) mean vegetation height. Points of interest in (a): (1)
downtown Tampa, (2) University of South Florida, (3) Tampa International Airport, (4) MacDill Air Force Base, (5) Lettuce Park a natural forest, and (6) a reserved
forest. Yellow circles are UHIs, while green circles are oases. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)

717
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Table 2 (Nowak et al., 1998). Surface roughness of urban landscape drives the
Multiple linear regression analysis results. microclimate fluctuations as it affects the airflow and wind speed
Coefficient Standardized coefficient (β) (Counehan, 1971; Gál and Sümeghy, 2007; Lin et al., 2017). Short trees
cast less shading effect on the nearby impervious surface or buildings
**
Constant 310.835 than tall trees, which can cover the roof and the windows to cool down
VF −3.008** −0.235
the temperature of buildings (Simpson and McPherson, 1998). Vege-
MVH −0.532** −0.452
tation height exerts an important influence on wind velocity which is
Model adjusted r2 = 0.347. critical in controlling local temperature fluctuations. While shrubs and
** p < 0.01. grass barely alter wind speed, the vertical distribution of tree canopy
does changes wind velocity (Geiger, 1958; Hall et al., 1958; Heisler,
with tree height and is optimized with a tree height of 20 m before 1990).
plateauing. The estimated marginal means of LST has a much higher Our results underscore the point that there are significant differ-
decreasing rate when mean vegetation height is less than 20 m, while ences in LST between the different levels of vegetation fraction and
height greater than 20 m has a decreasing rate close to zero (Fig. 7). vegetation height. Hitherto, our knowledge has been limited to the fact
that increasing vegetation cover leads to reduction of temperature.
However, our results show that this reduction may be more significant
5. Discussion
for vegetation cover above 93.33%. It is likely because very high ve-
getation cover (more than 93.33%) can effectively block most of the
Vegetation height engenders a significant effect on the magnitude of
solar radiation from reaching the impervious surfaces, which have high
LST. Building on the findings of Davis et al. (2016), this study expands
thermal conductivity that can increase the area of heating far beyond
our scientific understanding of the effects of vegetation vertical struc-
the point of incidence of sun’s radiation (Kodur, 2014). Studies have
ture on LST in urban landscapes by explicitly describing the quantita-
previously demonstrated this effect by showing that parks and reserved
tive relationship between LST and the vegetation height. The results
areas with a lot of vegetation have remarkable cooling effects (Cao
have important theoretical and management implications as they pro-
et al., 2010; Gaffin et al., 2008; Yu and Hien, 2006). In this study, we
vide urban planners and natural resource managers a better insight into
found that LST decreased steadily with increasing mean vegetation
increasing efficiency of UHI mitigation strategies through balancing the
height when height was less than 20 m. The decreasing rate was close to
lateral and vertical requirements of vegetation in the urban environ-
zero for mean vegetation height higher than 20 m. This result suggests
ment.
that mean vegetation height also affects LST and the cooling effect in-
creases with height. It is worth noting that vegetation height decreases
5.1. Theoretical implications LST at a constantly increasing rate and is optimized at 20 m before
plateauing, in which case the cooling effect of a vegetation patch with
While previous studies have examined the effect of vegetation cover mean height of 20 m would be similar to a vegetation patch with a
on LST (Chen et al., 2006; Weng et al., 2004; Yuan and Bauer, 2007), height of 25 m. This is possibly because very tall trees are those tree
little is known about the role of vegetation height on LST. In this study, species which gain height by compromising leaf area (e.g. palm tree
we compared the correlations between LST and vegetation height with and pine tree), which is one of the key controls on the capability of
that of cover. Results showed that vegetation height has a comparable evapotranspiration (Zhang et al., 2001). It is also worth noting that the
effect on decreasing LST as vegetation cover. A recent study by Davis correlation of vegetation structure could vary among cities of different
et al. (2016) suggested that vegetation volume (which considers both climate regions (Spronken-Smith and Oke, 1998). Different climatic
the lateral and vertical structure of vegetation) explains the variations conditions may significantly influence the amplitude of the effects of
of LST. In line with their result, our study highlights the importance of UHI. Meanwhile, vegetation structure, especially height, seems to be
the vertical structure of vegetation in UHI mitigation in urban land- different under different climatic conditions. For instance, trees in
scapes. The significant cooling effect of vegetation height is possibly Seattle Washington with a cold and wet climate are usually much taller
related to the influence of vegetation height on surface roughness of than those in Tampa that has a humid tropical weather. Therefore, the
urban landscape, shading effect, air dispersion, and wind velocity

Fig. 6. Estimated marginal means of LST for different levels of vegetation fraction.

718
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Fig. 7. Estimated marginal means of LST for different levels of mean vegetation height.

cooling effect of vegetation structure may not be consistent among ci- 6. Conclusions
ties (Zhou et al., 2017). Although vegetation height has a slightly
stronger correlation with LST than vegetation cover in the city of Urban Heat Island is a growing problem worldwide. Mitigation of
Tampa, the correlation could vary among cities in different climate UHI is necessary for cities to adapt to climate change and enhance
regions. A better understanding of the cooling efficiency of vegetation sustainable development at urban scale. Cooling cities with urban ve-
in urban areas could be achieved with future research that includes a getation management is a sustainable solution for urban heat mitiga-
comparison between cities from climatic regions. tion. Enhancing the cooling effect of urban vegetation requires a com-
Considerations should be made when applying the results of this prehensive understanding of how vegetation structure affects UHI. It
research to other cities. First, keeping different sources of data con- has been widely recognized that increasing vegetation cover can reduce
sistent in time is important. Although difference in the acquisition date land surface temperature in urban areas. However, the effect of vege-
of the LiDAR data and the land cover map and LST image may not have tation height on temperature has rarely been examined. Our study de-
much effect on our analysis due to the relatively small changes in ca- monstrated that vegetation height had a comparable effect on de-
nopy heights in Tampa throughout the year, using datasets acquired in creasing LST as vegetation cover. Although very high vegetation cover
the same time/season should be emphasized if possible. Since vegeta- can significantly reduce LST, it is extremely difficult to achieve such
tion changes in leaf area and canopy gap by seasons (Pettorelli et al., high vegetation cover in cities. Our results indicated that increasing
2005), canopy height obtained from LiDAR data acquired in winter may vegetation height has the potential to enhance cooling efficiency with
not accurately represent that in summer, especially in cities dominated the least amount of vegetation cover. Therefore, this study would
by deciduous trees. provide important insights into precise urban greenspace strategies to
mitigate UHI.

5.2. Management implications Acknowledgements

Numerous studies have demonstrated that areas dominated by large We thank the City of Tampa for their financial support of the City of
amount of vegetation such as nature reserves and parks can increase Tampa Urban Ecological
thermal comfort of their neighborhood (Cao et al., 2010; Spronken- Analysis 2006–2007 study, which supplied the high-resolution land
Smith and Oke, 1998; Yu and Hien, 2006). Our results are consistent cover data used in this study. We would also like to thank the Southwest
with those studies that very high vegetation cover is more efficient in Florida Water Management District for providing the LiDAR data used
UHI mitigation. However, gaining that amount of vegetation cover in in the study.
urban areas is extremely difficult for most cities. Even worse, instead of
gaining more green space, many cities are losing their vegetation as a References
result of urbanization (Nowak and Greenfield, 2012). How to augment
the cooling effect of vegetation in the very limited space in cities is thus Akbari, H., Pomerantz, M., Taha, H., 2001. Cool surfaces and shade trees to reduce energy
challenging for many urban planners. use and improve air quality in urban areas. Sol. Energy 70, 295–310. https://doi.org/
Our results showed that vegetation height has a significant impact 10.1016/S0038-092X(00)00089-X.
Alkama, R., Cescatti, A., 2016. Biophysical climate impacts of recent changes in global
on LST, which suggests that a city might need not only to just spread out
forest cover. Science 351, 600. https://doi.org/10.1126/science.aac8083.
green space, but also need to consider the height of vegetation. This Asner, G.P., Mascaro, J., 2014. Mapping tropical forest carbon: calibrating plot estimates
would be extremely important for compact cities which often have to a simple LiDAR metric. Remote Sens. Environ. 140, 614–624. https://doi.org/10.
limited space to expand their vegetation cover. At this point, although 1016/j.rse.2013.09.023.
Asner, G.P., Mascaro, J., Muller-Landau, H.C., Vieilledent, G., Vaudry, R., Rasamoelina,
there are plenty of species of trees, only a few make good trees in cities M., Hall, J.S., Breugel, M., 2011. A universal airborne LiDAR approach for tropical
(Pottstown Trees Inc, 2018). Tree species with high and large canopy forest carbon mapping. Oecologia 168, 1147–1160. https://doi.org/10.1007/
would be preferable in urban areas as they can provide shade for streets s00442-011-2165-z.
Barsi, J.A., Barker, J.L., & Schott, J.R. (2003). An atmospheric correction parameter
and buildings and reduce the temperature of the air below. Further- calculator for a single thermal band earth-sensing instrument. In, IGARSS 2003. 2003
more, maintaining and nurturing existing mature big trees would be IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE
more efficient and necessary. Cat. No.03CH37477) (pp. 3014-3016 vol.3015). doi: 10.1109/igarss.2003.1294665.

719
Q. Yu et al. Int J Appl Earth Obs Geoinformation 73 (2018) 712–720

Barsi, J.A., Schott, J.R., Palluconi, F.D., & Hook, S.J. (2005). Validation of a web-based For. Urban Green. 11, 21–30. https://doi.org/10.1016/j.ufug.2011.11.005.
atmospheric correction tool for single thermal band instruments. In Earth Observing Nowak, D.J., McHale, P.J., Ibarra, M., Crane, D., Stevens, J.C., Luley, C.J., 1998.
Systems X (Vol.5882, p. 58820E). International Society for Optics and Photonics. doi: Modeling the effects of urban vegetation on air pollution. Air Pollution Modeling and
10.1117/12.619990. Its Application XII. Springer, Boston, MA, pp. 399–407. https://doi.org/10.1007/
Bolund, P., Hunhammar, S., 1999. Ecosystem services in urban areas. Ecol. Econ. 29, 978-1-4757-9128-0_41.
293–301. https://doi.org/10.1016/S0921-8009(99)00013-0. Pettorelli, N., Vik, J.O., Mysterud, A., Gaillard, J.M., Tucker, C.J., Stenseth, N.C., 2005.
Campbell, S., 1996. Green cities, growing cities, just cities? Urban planning and the Using the satellite-derived NDVI to assess ecological responses to environmental
contradictions of sustainable development. J. Am. Plan. Assoc. 62, 296–312. https:// change. Trends Ecol. Evol. (Amst.) 20 (9), 503–510. https://doi.org/10.1016/j.tree.
doi.org/10.1080/01944369608975696. 2005.11.006.
Cao, X., Onishi, A., Chen, J., Imura, H., 2010. Quantifying the cool island intensity of Pottstown Trees Inc.10 best street trees. Data accessed on 5 March 2018. URL: https://
urban parks using ASTER and IKONOS data. Landsc. Urban Plan. 96, 224–231. www.pottstowntrees.org/H2-Best-street-trees.html.
https://doi.org/10.1016/j.landurbplan.2010.03.008. Simpson, J.R., McPherson, E.G., 1998. Simulation of tree shade impacts on residential
Carlson, T.N., Ripley, D.A., 1997. On the relation between NDVI, fractional vegetation energy use for space conditioning in Sacramento. Atmos. Environ. 32, 69–74. https://
cover, and leaf area index. Remote Sens. Environ. 62, 241–252. https://doi.org/10. doi.org/10.1016/s1352-2310(97)00181-7.
1016/s0034-4257(97)00104-1. Sobrino, J.A., Jiménez-Muñoz, J.C., Paolini, L., 2004. Land surface temperature retrieval
Chen, X.L., Zhao, H.M., Li, P.X., Yin, Z.Y., 2006. Remote sensing image-based analysis of from LANDSAT TM 5. Remote Sens. Environ. 90, 434–440. https://doi.org/10.1016/
the relationship between urban heat island and land use/cover changes. Remote Sens. j.rse.2004.02.003.
Environ. 104, 133–146. https://doi.org/10.1016/j.rse.2005.11.016. Sobrino, J.A., Jimenez-Munoz, J.C., Soria, G., Romaguera, M., Guanter, L., Moreno, J.,
Counehan, J., 1971. Wind tunnel determination of the roughness length as a function of Plaza, A., Martinez, P., 2008. Land surface emissivity retrieval from different VNIR
the fetch and the roughness density of three-dimensional roughness elements. Atmos. and TIR sensors. Ieee Trans. Geosci. Remote. Sens. 46, 316–327. https://doi.org/10.
Environ. 5 (8), 637–642. https://doi.org/10.1016/0004-6981(71)90120-x. 1967. 1109/tgrs.2007.904834.
Davis, A.Y., Jung, J., Pijanowski, B.C., Minor, E.S., 2016. Combined vegetation volume Spronken-Smith, R.A., Oke, T.R., 1998. The thermal regime of urban parks in two cities
and “greenness” affect urban air temperature. Appl. Geogr. 71, 106–114. https://doi. with different summer climates. Int. J. Remote Sens. 19, 2085–2104. https://doi.org/
org/10.1016/j.apgeog.2016.04.010. 10.1080/014311698214884.
Declet-Barreto, J., Knowlton, K., Jenerette, G.D., Buyantuev, A., 2016. Effects of urban Stone Jr, B., 2005. Urban heat and air pollution: an emerging role for planners in the
vegetation on mitigating exposure of vulnerable populations to excessive heat in climate change debate. J. Am. Plan. Assoc. 71, 13–25. https://doi.org/10.1080/
Cleveland, Ohio. Weather, Climate Soc. 8, 507–524. https://doi.org/10.1175/wcas- 01944360508976402.
d-15-0026.1. Su, Z., Schmugge, T., Kustas, W.P., Massman, W.J., 2001. An evaluation of two models for
Deilami, K., Kamruzzaman, M., Liu, Y., 2018. Urban heat island effect: A systematic re- estimation of the roughness height for heat transfer between the land surface and the
view of spatio-temporal factors, data, methods, and mitigation measures. Int. J. Appl. atmosphere. J. Appl. Meteorol. 40, 1933–1951. https://doi.org/10.1175/1520-
Earth Obs. Geoinf. 67, 30–42. https://doi.org/10.1016/j.jag.2017.12.009. 0450(2001)040<1933:aeotmf>2.0.co;2.
Estrada, F., Botzen, W.J.W., Tol, R.S.J., 2017. A global economic assessment of city po- Sud, Y.C., Shukla, J., Mintz, Y., 1988. Influence of land surface roughness on atmospheric
licies to reduce climate change impacts. Nat. Clim. Chang. 7, 403. https://doi.org/10. circulation and precipitation: a sensitivity study with a general circulation model. J.
1038/nclimate3301. Appl. Meteorol. 27, 1036–1054. https://doi.org/10.1175/1520-0450(1988)
Fu, P., Weng, Q., 2016. A time series analysis of urbanization induced land use and land 027<1036:iolsro>2.0.co;2.
cover change and its impact on land surface temperature with Landsat imagery. Susca, T., Gaffin, S.R., Dell’Osso, G.R., 2011. Positive effects of vegetation: urban heat
Remote Sens. Environ. 175, 205–214. https://doi.org/10.1016/j.rse.2015.12.040. island and green roofs. Environ. Pollut. 159, 2119–2126. https://doi.org/10.1016/j.
Gaffin, S.R., Rosenzweig, C., Khanbilvardi, R., Parshall, L., Mahani, S., Glickman, H., envpol.2011.03.007.
Goldberg, R., Blake, R., Slosberg, R.B., Hillel, D., 2008. Variations in New York city’s Tan, J., Zheng, Y., Tang, X., Guo, C., Li, L., Song, G., Zhen, X., Yuan, D., Kalkstein, A.J., Li,
urban heat island strength over time and space. Theor. Appl. Climatol. 94, 1–11. F., 2010. The urban heat island and its impact on heat waves and human health in
https://doi.org/10.1007/s00704-007-0368-3. Shanghai. Int. J. Biometeorol. 54, 75–84. https://doi.org/10.1007/s00484-009-
Gál, T.M., Sümeghy, Z., 2007. Mapping the roughness parameters in a large urban area 0256-x.
for urban climate applications. Acta Climatologica ET Chorologica 40, 27–36. Weng, Q., Fu, P., 2014. Modeling annual parameters of clear-sky land surface temperature
Geier-Hayes, K., Hayes, M.A., & Basford, D.D. (1995). Determining individual tree shade variations and evaluating the impact of cloud cover using time series of Landsat TIR
length: A guide for silviculturists. Forest Service general technical report. In: Forest data. Remote Sens. Environ. 140, 267–278. https://doi.org/10.1016/j.rse.2013.09.
Service, Ogden, UT (United States). Intermountain Research Station. doi: 10.5962/ 002.
bhl.title.100476. Weng, Q., Lu, D., Schubring, J., 2004. Estimation of land surface temperature–vegetation
Geiger, R., 1958. The modification of microclimate by vegetation in open country and in fraction relationship for urban heat island studies. Remote Sens. Environ. 89,
hilly country. Climatology and Microclimatology: Proceedings of the Canberra 467–483. https://doi.org/10.1016/j.rse.2003.11.005.
Symposium. United Nations Educational, Scientific and Cultural Organization, Paris, Weng, Q.H., Lu, D.S., Liang, B.Q., 2006. Urban surface biophysical descriptors and land
France, pp. 255–258. surface temperature variations. Photogramm. Eng. Remote Sensing 72, 1275. https://
Hakner, J. (2017). “Heat island” effect could double climate change costs for world’s doi.org/10.14358/pers.72.11.1275.
cities. University of Sussex, updated by 29 May 2017. Data accessed on 15 March White, M.A., Nemani, R.R., Thornton, P.E., Running, S.W., 2002. Satellite evidence of
2018. URL: https://www.sciencedaily.com/releases/2017/05/170529133714.htm. phenological differences between urbanized and rural areas of the eastern United
Hall, N.; Russel, V.S.; & Hamilton, C.D. (1958). Trees and Microclimate. Pages 259–264 in States deciduous broadleaf forest. Ecosystems 5, 260–273. https://doi.org/10.1007/
Climatology and Microclimatology: proceedings of the Canberra Symposium. United s10021-001-0070-8.
Nations Educational, Scientific and Cultural Organization, Paris, France. Soil texture, Wu, H., Ye, L.P., Shi, W.Z., Clarke, K.C., 2014. Assessing the effects of land use spatial
topography, and canopy effects on microclimate in scrub oak thickets on Martha’s structure on urban heat islands using HJ-1B remote sensing imagery in Wuhan,
Vineyard Geiger. China. Int. J. Appl. Earth Obs. Geoinf. 32, 67–78. https://doi.org/10.1016/j.jag.
Heisler, G.M., 1990. Mean wind speed below building height in residential neighborhoods 2014.03.019.
with different tree densities. ASHRAE Trans. 96 (1), 1389–1396. Xian, G., Crane, M., 2003. Impacts of urban development on seagrass in tampa Bay.
Jiménez‐Muñoz, J.C., Sobrino, J.A., 2006. Error sources on the land surface temperature Proceedings of the 19th National Environmental Monitoring Conference.
retrieved from thermal infrared single channel remote sensing data. Int. J. Remote Xian, G., Crane, M., 2005. Assessments of urban growth in the Tampa Bay watershed
Sens. 27, 999–1014. https://doi.org/10.1080/01431160500075907. using remote sensing data. Remote Sens. Environ. 97, 203–215. https://doi.org/10.
Johnston, I. (2017). Urban “heat island” effect could intensify climate change, making 1016/j.rse.2005.04.017.
cities up to 7C warmer. Independent, published by 29 May 2017. Data accessed on 15 Yan, H., Fan, S., Guo, C., Wu, F., Zhang, N., Dong, L., 2014. Assessing the effects of
March 2018. URL: https://www.independent.co.uk/environment/urban-heat-island- landscape design parameters on intra-urban air temperature variability: the case of
cities-climate-change-worse-global-warming-7-degrees-cool-roofs-pavements- Beijing, China. Buil. Environ. 44. https://doi.org/10.1016/j.buildenv.2014.03.007.
a7761846.html. Yu, C., Hien, W.N., 2006. Thermal benefits of city parks. Energy Build. 38, 105–120.
Kodur, V., 2014. Properties of concrete at elevated temperatures. Isrn Civ. Eng. 2014. https://doi.org/10.1016/j.enbuild.2005.04.003.
https://doi.org/10.1155/2014/468510. Yuan, F., Bauer, M.E., 2007. Comparison of impervious surface area and normalized
Landry, S.M., Pu, R., 2010. The impact of land development regulation on residential tree difference vegetation index as indicators of surface urban heat island effects in
cover: an empirical evaluation using high-resolution IKONOS imagery. Landsc. Urban Landsat imagery. Remote Sens. Environ. 106, 375–386. https://doi.org/10.1016/j.
Plan. 94 (2), 94–104. https://doi.org/10.1016/j.landurbplan.2009.08.003. rse.2006.09.003.
Lin, P., Gou, Z., Lau, S.S.-Y., Qin, H., 2017. The impact of urban design descriptors on Zhang, L., Dawes, W.R., Walker, G.R., 2001. Response of mean annual evapotranspiration
outdoor thermal environment: a Literature Review. Energies 10, 2151. https://doi. to vegetation changes at catchment scale. Water Resour. Res. 37 (3), 701–708.
org/10.3390/en10122151. https://doi.org/10.1029/2000wr900325.
Maurer, K.D., Bohrer, G., Kenny, W.T., Ivanov, V.Y., 2015. Large-eddy simulations of Zhou, W., Huang, G., Cadenasso, M.L., 2011. Does spatial configuration matter?
surface roughness parameter sensitivity to canopy-structure characteristics. Understanding the effects of land cover pattern on land surface temperature in urban
Biogeosciences 12, 2533–2548. https://doi.org/10.5194/bg-12-2533-2015. landscapes. Landsc. Urban Plan. 102, 54–63. https://doi.org/10.1016/j.landurbplan.
Murgante, B., Borruso, G., Lapucci, A., 2011. Sustainable development: concepts and 2011.03.009.
methods for its application in urban and environmental planning. Geocomputation, Zhou, W., Wang, J., Cadenasso, M.L., 2017. Effects of the spatial configuration of trees on
Sustainability and Environmental Planning. Springer, pp. 1–15. https://doi.org/10. urban heat mitigation: a comparative study. Remote Sens. Environ. 195, 1–12.
1007/978-3-642-19733-8_1. https://doi.org/10.1016/j.rse.2017.03.043.
Nowak, D.J., Greenfield, E.J., 2012. Tree and impervious cover change in US cities. Urban

720